Inspired by an audiobook project created earlier this semester, we want to build a web-based tool that generates multiple assemblages of the three texts used in “The Yellow-Wallpapered Box Social Story of an Hour” (“The Yellow Wallpaper” by Charlotte Perkins Gilman, “The Story of an Hour” by Kate Chopin, and “The Box Social” by James Reaney). Through a series of programmatic text manipulation experiments, both more and less aleatoric, we hope to address the following question: are arbitrary, systematic text manipulations, both intratext and intertext, meaningful for interpretation and/or scholarly research methods, and at what levels of automation might the usefulness break down? The scholarly inspiration for this endeavor lies strongly with “Deformance and Interpretation.” Beyond the support from Samuels and McGann, we can also take practical reinforcements from visual media scholars that run with theoretical implications of deformance. CUNY’s own Kevin L. Ferguson has demonstrated the usefulness of creating image composites from movie stills in the aid of film studies with digital technologies. Similarly, Jason Mittel’s “Deformin’ in the Rain” collection provides fertile ground for how a curated set of deformances can be in delivering a fresh perspective on a text. As Mittel summarizes elsewhere “[deformance] strives to make the original work strange in some unexpected way, deforming it unconventionally to reveal aspects that are conventionally obscured in its normal version and discovering something new from it.”
We aim to explore a few different experimental methods. We may also create an interactive experience where a visitor could refresh the page or an element and remix the interwoven stories into a new text paragraph by paragraph. Another might operate as a sort of literary Mad Libs, taking parts of speech from one text and applying them to another. Inspired by erasure poetry, we may also explore removing portions of text from the stories (see also this ErasureMaker project). Through our experimentation we hope to explore such questions as: Is the whole of a story, or the layering of multiple stories, more or less than the sum of its parts? Is it possible (especially as someone other than the author) to distill a story down to its essence? Can you tell a coherent story without all the parts of speech? How much of a story can you change or take away before it becomes unrecognizable? How much fidelity, if any, do we owe the original texts and their authors?
To the extent possible, we hope that all work can be done in a web browser without pre-processed work offline or a server side component. We have experimented with a natural language processing (NLP) library written in Nodejs called compromise. This library has much to recommend it for our purposes. The text parsing, filtering, and tagging methods are varied, but kept relatively simple and performant. The library doesn’t focus on the statistical methods and models common in NLP packages. Nor is it the most accurate tool for NLP. Rather, it tries its best to treat text as data. This emphasis sometimes results in less than accurate outputs, which helps the process unfold systematically, but somewhat arbitrarily. Other benefits include the ability to use the library locally via content delivery network (CDN), letting us perform our text mangling in the browser, avoiding server side development, and granting us free hosting of the website via Github Pages. A rough and ready exercise to determine compromise’s appropriateness for our experiments is available here. In this web page, we’ve ingested the text as a string literal, tokenized it, then broke the text into some typical parts of speech kept in sequential order. Of particular interest are perfectly imperfect “nouns” and “verbs,” which are really phrases or groups of the parts of speech that are ripe for remixing. Some examples include the following:
Nouns
- the open country
- that beautiful door
- john dear
- that long smooch
- my wellhidden ropeyou
- separate little houses
- coheirs anyhow
Verbs
- will proudly declare
- never saw such ravages
- outlines run off
- suddenly commit
In this proof of concept iteration of the project, we are limiting ourselves to the three stories we worked with on the audiobook project as we are already very familiar with these texts and know that intentionally layering them together yields exciting results. As such we feel confident in our ability to judge how much the use of these digital tools in remixing these texts has the potential to add or distract from the discourse, whether the new texts our experiments yield create a dialogue or conflict with the originals, and what we can learn from each.
Working Bibliography
Ferguson, Kevin L. “Digital Surrealism: Visualizing Walt Disney Animation Studios.” DHQ: Digital Humanities Quarterly: Digital Surrealism: Visualizing Walt Disney Animation Studios, 2017, http://www.digitalhumanities.org/dhq/vol/11/1/000276/000276.html.
Mittel, Jason. “Deformin’ in the Rain.” Deformin’ in the Rain on Vimeo, 2017, https://vimeo.com/showcase/6603776.
Mittel, Jason. “Videographic Criticism as a Digital Humanities Method.” Debates in the Digital Humanities, 2019, https://dhdebates.gc.cuny.edu/read/4805e692-0823-4073-b431-5a684250a82d/section/b6dea70a-9940-497e-b7c5-930126fbd180/resource/ec709ed8-8ce2-4383-969b-2a8ad1887823.
Samuels, Lisa and Jerome J McGann. “Deformance and Interpretation.” New Literary History, vol. 30 no. 1, 1999, p. 25-56. Project MUSE, doi:10.1353/nlh.1999.0010.
